Short answer
First find out what interferes and where from. Then try fixes starting with the quickest and cheapest: cabling, grounding, ferrites and filters, firmware. Only when that is not enough, change the enclosure or the board design. Retest after every change.
Find the cause first
Randomly adding ferrites sometimes helps, but often just moves the problem elsewhere. The test report and a few simple experiments usually show where to look:
- Narrow peaks at regular intervals are usually multiples (harmonics) of some clock. Find which crystal or bus they correspond to.
- A broad “hill” in the spectrum is often caused by a switch-mode power supply or a DC/DC converter.
- Conducted emissions (up to 30 MHz) usually point to the power input and its filtering.
- If a peak changes when a cable is disconnected or rerouted, that cable is the main radiator.
- For immunity (ESD, fast transients, surges), what matters is where the disturbance enters and what exactly fails: a reset, a communication dropout or damage.
We then pinpoint the exact spot on the board with a near-field probe. It shows which component or trace radiates the most.
Fixes from the cheapest
The order is indicative. Only testing shows what helps your particular device.
1. Cabling
effort: lowCables are the most common antenna. Current flowing along them in the same direction there and back (common-mode current) radiates even when the device itself is fine. A cable also brings interference inside, for example during the fast transient test.
- Shorten cables and route them close to the chassis, not loosely in the air.
- Use twisted pairs for signals and power.
- Terminate the cable shield around the full circumference of the connector, not with a pigtail.
- Separate power and signal wires; do not run them in one bundle.
2. Chassis bonding and grounding
effort: lowShielding and filters only work if they have somewhere to divert the interference. A long, thin wire to the chassis has high impedance at high frequencies and hardly helps.
- Bond to the chassis short and wide (strap, direct contact), not with a long wire.
- Ensure conductive contact between the cover and the enclosure; remove paint at contact points.
- Connect metal connectors and their shields directly to the chassis.
- Check where the electronics ground connects to the chassis and whether unnecessary loops are formed.
3. Ferrites and filters
effort: low to mediumA ferrite ring on a cable suppresses common-mode current. It usually helps in the range of tens to hundreds of MHz. A filter on the power input reduces conducted emissions from switch-mode supplies and improves immunity.
- Ferrite rings or clamp-on ferrites on the cables where peaks showed up.
- An AC or DC filter on the power input.
- Fill capacitors and chokes into unpopulated positions on the board.
- For immunity, add surge and ESD protection (e.g. TVS diodes) on the interfaces.
4. Firmware tuning
effort: mediumA lot of interference comes from clocks and fast signal edges. Some of it can be reduced by configuration alone, without touching the hardware. Firmware also helps with immunity: the device recovers by itself after a disturbance.
- Enable spread-spectrum clocking if the interfaces tolerate it.
- Slow down output edges (GPIO and bus slew-rate settings).
- Disable unused peripherals, clock outputs and buses.
- Move switching and communication frequencies out of problematic bands.
- For immunity: watchdog, communication checks and retries, safe return to operation.
5. Enclosure and shielding
effort: medium to higherA metal enclosure shields only as well as its parts are bonded together. Long slots and seams radiate like antennas themselves. A plastic enclosure does not shield at all.
- Conductive gaskets or spring contacts between the cover and the enclosure, more screws.
- Several small openings instead of one long slot.
- A shielding can over the problematic part of the board.
- For plastic, a conductive coating inside the housing or internal shielding.
6. Electronics design
effort: highWhen the board or the circuit is the cause, other measures only mitigate the consequences. A design change means a new board revision, but it solves the problem at the root.
- A PCB stackup with a solid ground plane under fast signals.
- Short current loops in switch-mode supplies and decoupling capacitors.
- Placement: fast circuits away from connectors and board edges.
- Filtering and protection right at the connectors.
- Choosing components with slower edges or lower emissions.
How we handle it
- We measure the device in our EMC chamber and show you where and by how much it exceeds the limit.
- We look for the cause right at the chamber, together with you: rerouting cables, near-field probing, disconnecting parts.
- We try fixes on site (ferrites, bonding, temporary shielding, firmware settings) and retest immediately.
- You get the result with before-and-after charts and a recommendation on what to implement in production.
When the problem lies in the board or enclosure design, our engineering team proposes a fix and we retest the next revision right away. We can usually offer testing within a week.
How to prevent it next time
- Think about EMC at the design stage: board stackup, ground plane, filtering at the connectors, enclosure bonding.
- Do pre-compliance testing on the prototype while changes are cheap.
- Prepare for testing with our checklist, so chamber time is not wasted.
The complete CE guide as a PDF
All 11 guides in one document: directives, EMC and safety testing, technical documentation, manual and rating label. For printing and for your team.
Did your device fail?
Get an indicative price for pre-compliance testing and root-cause analysis. We reply within 24 hours.